mirrors/pulseaudio
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/pulseaudio/pulseaudio.git synced 2025-11-09 13:29:59 -05:00
Code Issues Projects Releases Packages Wiki Activity Actions

sconv: avoid multiply in ARM NEON float->s16 conversion

Browse source 
optimization idea taken from libavresample

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
This commit is contained in:
Peter Meerwald 2013-07-02 11:43:35 +02:00 • committed by Tanu Kaskinen
parent 0e8d26a6a1
commit 40c3c117e7
1 changed files with 3 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

9
src/pulsecore/sconv_neon.c
View file

@ -36,12 +36,9 @@ static void pa_sconv_s16le_from_f32ne_neon(unsigned n, const float *src, int16_t
"movs %[n], %[n], lsr #2 \n\t" "movs %[n], %[n], lsr #2 \n\t"
"beq 2f \n\t" "beq 2f \n\t"
"vdup.f32 q1, %[scale] \n\t"
"1: \n\t" "1: \n\t"
"vld1.32 {q0}, [%[src]]! \n\t" "vld1.32 {q0}, [%[src]]! \n\t"
"vmul.f32 q0, q0, q1 \n\t" /* scale */ "vcvt.s32.f32 q0, q0, #31 \n\t" /* s32<-f32 as 16:16 fixed-point, with implicit multiplication by 32768 */
"vcvt.s32.f32 q0, q0, #16 \n\t" /* s32<-f32 as 16:16 fixed-point */
"vqrshrn.s32 d0, q0, #16 \n\t" /* shift, round, narrow */ "vqrshrn.s32 d0, q0, #16 \n\t" /* shift, round, narrow */
"subs %[n], %[n], #1 \n\t" "subs %[n], %[n], #1 \n\t"
"vst1.16 {d0}, [%[dst]]! \n\t" "vst1.16 {d0}, [%[dst]]! \n\t"
@ -50,8 +47,8 @@ static void pa_sconv_s16le_from_f32ne_neon(unsigned n, const float *src, int16_t
"2: \n\t" "2: \n\t"
: [dst] "+r" (dst), [src] "+r" (src), [n] "+r" (n) /* output operands (or input operands that get modified) */ : [dst] "+r" (dst), [src] "+r" (src), [n] "+r" (n) /* output operands (or input operands that get modified) */
: [scale] "r" (32768.0f) /* input operands */ : /* input operands */
: "memory", "cc", "q0", "q1" /* clobber list */ : "memory", "cc", "q0" /* clobber list */
); );
/* leftovers */ /* leftovers */